<div><div dir="auto">I think it is. The way th calls are set for performance reasons seems to be causing this</div></div><div dir="auto">Ideally we should ensure to use right dialect for  tls calls I think musl is exposing a bug in Mesa </div><div dir="auto">Which we covered by disabling tls in Glx once we disable that then this problem pops up so we disable ASM </div><div dir="auto"><br></div><div dir="auto">So this is a workaround until then </div><div><br><div class="gmail_quote"><div dir="ltr">On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross &lt;<a href="mailto:ross.burton@intel.com">ross.burton@intel.com</a>&gt; wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Isn&#39;t this a bit overkill?  Is the assembler actually broken?<br>
<br>
Ross<br>
On Tue, 2 Oct 2018 at 20:21, Khem Raj &lt;<a href="mailto:raj.khem@gmail.com" target="_blank">raj.khem@gmail.com</a>&gt; wrote:<br>
&gt;<br>
&gt; some unknown part of mesa&#39;s x86 assembly<br>
&gt; code is broken by readonly text segments<br>
&gt;<br>
&gt; [ YOCTO #12918 ]<br>
&gt;<br>
&gt; Signed-off-by: Khem Raj &lt;<a href="mailto:raj.khem@gmail.com" target="_blank">raj.khem@gmail.com</a>&gt;<br>
&gt; ---<br>
&gt;  meta/recipes-graphics/mesa/mesa.inc | 2 +-<br>
&gt;  1 file changed, 1 insertion(+), 1 deletion(-)<br>
&gt;<br>
&gt; diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc<br>
&gt; index 8d0e2cb67c..e6f3712706 100644<br>
&gt; --- a/meta/recipes-graphics/mesa/mesa.inc<br>
&gt; +++ b/meta/recipes-graphics/mesa/mesa.inc<br>
&gt; @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = &quot;--enable-${OSMESA},--disable-${OSMESA}&quot;<br>
&gt;  PACKAGECONFIG[unwind] = &quot;--enable-libunwind,--disable-libunwind,libunwind&quot;<br>
&gt;<br>
&gt;  EXTRA_OECONF_remove_libc-musl = &quot;--enable-glx-tls&quot;<br>
&gt; -EXTRA_OECONF_append_libc-musl = &quot; --disable-glx-tls&quot;<br>
&gt; +EXTRA_OECONF_append_libc-musl = &quot; --disable-asm --disable-glx-tls&quot;<br>
&gt;  # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2)<br>
&gt;  FULL_OPTIMIZATION_append = &quot; -fno-omit-frame-pointer&quot;<br>
&gt;<br>
&gt; --<br>
&gt; 2.19.0<br>
&gt;<br>
&gt; --<br>
&gt; _______________________________________________<br>
&gt; Openembedded-core mailing list<br>
&gt; <a href="mailto:Openembedded-core@lists.openembedded.org" target="_blank">Openembedded-core@lists.openembedded.org</a><br>
&gt; <a href="http://lists.openembedded.org/mailman/listinfo/openembedded-core" rel="noreferrer" target="_blank">http://lists.openembedded.org/mailman/listinfo/openembedded-core</a><br>
</blockquote></div></div>